Sand castings are inspected with six NDT methods, and the one to call out depends on where the defect you are guarding against sits: visual, liquid penetrant and magnetic particle testing find what breaks the surface; radiography, ultrasonics and CT find what is buried in the section. No single method covers every defect, and the ASTM casting grades do not choose one for you — beyond a visual check, every test has to be ordered separately.
Common NDT Methods for Sand Castings
Six methods cover sand castings between them: three that read the surface (VT, PT, MT) and three that see through the section (RT, UT, CT).
| NDT method | Best for detecting | Typical materials | Main limitation |
|---|---|---|---|
| Visual Testing (VT) | Surface defects, mismatch, cracks, poor finish | All cast metals | Cannot detect internal defects |
| Liquid Penetrant Testing (PT) | Surface-breaking cracks and porosity | Aluminum, steel, stainless steel, bronze | Surface must be reasonably clean and smooth |
| Magnetic Particle Testing (MT) | Surface and near-surface cracks | Ferromagnetic steels and irons | Cannot be used on aluminum or most non-ferrous alloys |
| Radiographic Testing (RT) | Porosity, shrinkage, inclusions, internal cavities | Most casting alloys | Higher cost and radiation controls |
| Ultrasonic Testing (UT) | Internal cracks, inclusions, wall discontinuities | Steel, iron, some aluminum castings | Rough surfaces and coarse grain can reduce accuracy |
| Computed Tomography (CT) | Detailed 3D internal defects | Most casting materials | High equipment cost |
1. Visual Testing
Visual testing is the one inspection every casting gets and the only one included in the base ASTM grade. Order a valve body to A216 WCB and the specification already requires a visual check that the casting is free of sand, scale and cracks; a crack is rejectable under every ASTM casting grade without any supplementary requirement. Everything past that check is optional and has to be ordered.
What VT settles beyond cracks is surface condition: parting-line mismatch, surface texture, pinholes, sand expansion scabs, and how cut-off, welded and machined areas were dressed.
2. Liquid Penetrant Testing
Liquid penetrant testing finds whatever is open to the surface — cracks, hot tears, cold shuts, surface shrinkage, porosity that reaches the skin — on any nonporous metal, which makes it the surface method for the alloys MT cannot touch: austenitic stainless such as CF8 and CF8M. On carbon steel and iron it works as well, but MT is faster and reads slightly below the surface, so PT on a magnetic grade is usually chosen only for a finished machined face.
The table’s “reasonably clean and smooth” is two conditions of very different weight. Clean is mandatory: penetrant will not enter a flaw through scale, oil or burnt-on sand, so the area has to be bare and dry. Smooth is conditional: an as-cast surface is acceptable, and grinding is called for only where roughness would itself hold penetrant and mask an indication.
PT reports the presence, position and rough extent of a flaw, not its depth, so a PT callout is a surface-quality requirement, not a soundness one.
3. Magnetic Particle Testing
Magnetic particle testing is the surface method we run on carbon steel, low-alloy steel, gray iron, ductile iron and the martensitic and duplex stainless grades — everything ferromagnetic — and on those alloys it does one thing PT cannot: it finds flaws just under the surface as well as at it. A crack or hot tear under a thin skin of metal still breaks the flux path and collects particles; penetrant never reaches it.
The limitation cell says aluminum, but the exclusion that matters in a ferrous foundry is austenitic stainless. CF8, CF8M and the other austenitic grades are non-magnetic, so there is no flux to leak and MT gives no indication at all; those castings go to PT. Duplex grades are ferromagnetic and take MT normally.
4. Radiographic Testing
Radiography is the internal-soundness method for sections up to a few inches: gas porosity, shrinkage and sand or slag inclusions show as density differences on film, and ASTM grades them against reference radiographs so a callout can name a severity level. Cracks and hot tears are planar, show only when the beam runs roughly along their plane, and appear in the reference radiographs as one ungraded illustration each — which is why the defect table gives RT a full mark on volumetric defects and a weaker one on cracks.
The reference radiograph set is chosen by section thickness, and the callout must name it:
| Reference radiographs | Section thickness |
|---|---|
| ASTM E446 | up to 2 in. (50.8 mm) |
| ASTM E186 | 2 to 4½ in. (50.8 to 114 mm) |
| ASTM E280 | 4½ to 12 in. (114 to 305 mm) |
ASTM E689 extends all three to ductile iron.
Cost climbs with thickness: sections under 2 in. can be shot with a medium-voltage X-ray tube, while the E280 band assumes cobalt-60 or a 2 MV-and-up machine, longer exposures and a licensed bay — by which point UT is usually the cheaper answer.
5. Ultrasonic Testing
Ultrasonic testing is the internal method for thick steel sections and for anything radiography cannot reach economically: a pulse-echo beam sent into the casting returns an echo from any reflector — a shrinkage cavity, an internal crack, an inclusion — and from the back wall, and the timing and amplitude of that echo give depth and rough size in one reading. The ASTM practice is A609, and its scope is narrower than the table’s “steel, iron, some aluminum”: heat-treated carbon, low-alloy and martensitic stainless steel castings only.
Two things limit it on sand castings. An as-cast surface scatters the beam, so scan areas are usually ground or machined first, which is why UT often goes after rough machining rather than on the raw casting. And coarse or graphite-bearing structures scatter it internally: austenitic stainless with its coarse as-cast grain sits outside A609, and gray iron is close to untestable for flaws.
Gray iron deserves numbers, because “iron” in the table hides the difference between the two irons we pour. In a published nodularity study, longitudinal velocity ran 5,643 m/s at 90 % nodularity and 3,948 m/s in flake-graphite gray iron; the flakes reflect and scatter the wave, attenuation climbs, and a real flaw echo disappears into the noise. Ductile iron at good nodularity behaves close to steel and can be flaw-scanned.
The velocity drop from nodular to flake graphite also makes UT useful on ductile iron in a second way: a velocity reading on a machined pad checks nodularity without cutting a sample, and a spec can set an accept/reject velocity for it. That is a metallurgical check, not an A609 flaw scan, and the order should name which one it wants.
6. Computed Tomography
Computed tomography gives a full 3D map of the interior — every pore, cavity and inclusion located and sized, with no overlapping structures to interpret — and for a medium-to-large ferrous sand casting it is almost never the practical choice, because a laboratory CT tube gets through only a few centimetres of steel. A published CT study on micro-focus systems put the usable steel path at about 20 mm with a 200 kV tube and 30 mm at 320 kV.
Where CT earns its cost on ferrous work is small parts and first articles: a thin-walled investment-cast stainless part, or a first-off casting small enough to scan where you want the actual cavity shape rather than a film shadow before committing the pattern.
Which NDT Method Should You Use?
Pick by where the defect lives and what the alloy allows. Surface flaws: MT on any magnetic grade, PT on austenitic stainless. Internal soundness: RT where the section is under a few inches, UT where it is thicker or the scan areas will be machined anyway. CT only when the part is small enough to scan and the question is worth a lab day.
The second half of the decision is the one drawings most often skip: none of this comes with the grade. The common-requirements specifications — A703 for pressure-containing steel, A781 for general-industrial steel, A834 for iron — carry every test beyond visual as a supplementary requirement that applies “only when specified individually by the purchaser in the order”. The Steel Founders’ Society’s ordering guide maps them for A703: MT is S4, RT is S5, PT is S6, UT is S7. Leave them off the PO and a fully grade-compliant casting arrives with a visual check and nothing else.
Naming a method is still not a complete order. It needs the zones (the pressure wall and the flange faces, not the whole casting), the reference standard with a severity or acceptance level, and the stage — MT or PT after heat treatment catches cracks that open in the quench and that a check on the raw casting cannot see, and UT after rough machining gets the surface it needs.
NDT Method vs. Typical Sand Casting Defects
Surface-breaking defects — cracks, hot tears, cold shuts — are a PT or MT job, and buried ones — gas porosity, shrinkage, sand inclusions, internal cracks — are an RT or UT job. The table below marks it per defect for the five production methods: ✓✓ the method to rely on, ✓ one that finds the defect under normal conditions, “Sometimes” one that depends on the flaw’s orientation or depth, and — one not to count on.
| Casting defect | VT | PT | MT | RT | UT |
|---|---|---|---|---|---|
| Surface crack | ✓ | ✓✓ | ✓✓ | Sometimes | Sometimes |
| Hot tear | ✓ | ✓✓ | ✓✓ | ✓ | ✓ |
| Gas porosity | Surface only | Surface-connected | Limited | ✓✓ | ✓ |
| Shrinkage cavity | Sometimes | — | — | ✓✓ | ✓✓ |
| Sand inclusion | Surface only | — | — | ✓✓ | ✓ |
| Cold shut | ✓ | ✓ | ✓* | ✓ | Sometimes |
| Internal crack | — | — | Limited | ✓ | ✓✓ |
Gas porosity is the row that catches people: PT sees only the pores that reach the skin, RT sees the population under it, so a boss that has to be pressure-tight needs the RT mark, not the PT one.
Cold shut is the one row with a mark in every column, because it is a surface-breaking seam of unfused metal. Its MT mark holds only on magnetic grades — on CF8M a cold shut is a PT job — and that condition applies to every entry in the MT column.
Conclusion
Surface defects are a VT, MT or PT question; internal soundness is an RT or UT question; CT is a laboratory tool for parts small enough to fit in one. The alloy narrows it further — austenitic stainless takes PT not MT, gray iron does not take UT — and the ASTM grade settles none of it beyond a visual check. Put the method, the zone and the acceptance level on the drawing and the NDT question is closed before the pattern is cut.